How to Build a Lead Scoring Model (Free Template and Example)

To build a lead scoring model, list what your best customers have in common, give points for fit and for buying intent, and set three score bands that each trigger one action. Then test the model against your past deals and adjust it every quarter. A simple spreadsheet or your CRM's scoring tool is enough to start.

This guide walks through the seven steps, with a copyable template and a worked example for a small B2B team.

What is a lead scoring model?

A lead scoring model is a set of rules that gives each lead a number, so your team knows who to contact first. Most models measure two things:

• Fit: does this lead look like your customer? Think job title, company size, industry and region.
• Intent: is this lead showing interest now? Think a demo request, a pricing page visit or a reply to an email.

The two answer different questions, which is why good models keep them apart. A VP at a perfect-fit company who has never heard of you is a prospect to work. A student who downloaded three guides is active, but will never buy. Add those scores into one number and the two can look the same.

Scoring also differs from lead qualification. Scoring is an automatic guess from data, made before anyone talks to the lead. Qualification is a person confirming the facts in conversation. Scoring decides the order of the queue, and qualification decides who gets a meeting.

Do you need a lead scoring model yet?

Not always. Scoring solves a prioritisation problem, so it only helps if you have one.

Your situation	What to do
Your team reaches every lead within a day	Skip scoring and call everyone
Leads pile up and reps pick the ones that look good	Build the simple model below
Good leads wait behind poor ones	Build the model and tie it to routing
You have thousands of leads and a long conversion history	Try predictive scoring in your CRM

Predictive tools need data most small teams do not have. Salesforce's Einstein Lead Scoring needs at least 1,000 leads created in the last 200 days, with at least 120 of them converted, before it can build a model on your own data. Below that, it falls back to a global model built from anonymised data across many customers. A rules-based model is the honest choice for most small teams, because you can read every rule and explain why a lead scored what it did.

How to build a lead scoring model in 7 steps

1. Pull your last 20 to 50 deals

Export your recent closed-won and closed-lost deals. Include the lost ones, or you will end up describing everyone who talks to you, not the people who buy. With fewer deals, use what you have and treat the first version as a hypothesis.

2. Find what separates wins from losses

For each deal, note the job title, company size, industry, lead source and the actions taken before the first call. Look for traits that show up far more often in wins than in losses. A trait common to both tells you nothing, even if it feels important.

3. Pick 4 to 6 fit criteria

Choose the criteria that separated your deals and that you can fill in reliably for most leads. If half your leads have no company size, either enrich that field or drop it. Your ideal customer profile, the same one you use for prospect lists, is a good starting point.

4. Pick 3 to 5 intent signals

Choose actions that show buying interest. Weight a demo request far above a newsletter signup. Be careful with email opens, because Apple Mail Privacy Protection and security scanners can trigger them without anyone reading.

5. Add negative points

Take points off for clear bad-fit signals: a competitor's domain, a job applicant, a personal email address if you only sell to companies, or a region you do not serve. Without this step, active leads who will never buy can climb to the top of the queue.

6. Set three bands and one action for each

Reps do not need a grading system with ten levels. They need to know what to do next. Agree the action with whoever works the leads before you switch anything on.

7. Test it on old deals, then review it quarterly

Run the model across deals you have already closed. If won deals do not cluster in your top band, your points are wrong. After launch, review the results every quarter, because your market, messaging and channels will change.

Lead scoring template

Copy this into a spreadsheet and replace the criteria with your own. The points are a starting example, not a benchmark, and your own deal data should change them.

Fit score (out of 50)

Criterion	Match	Points
Job title	Decision-maker	15
Job title	Influencer or manager	8
Company size	Your core range	15
Company size	Adjacent range	5
Industry	Target industry	10
Region	Region you serve	5
Tech or signal	Uses a key tool or matches a buying signal	5

Intent score (capped at 50)

Signal	Points
Requested a demo or call	30
Replied positively to outreach	25
Visited the pricing page	10
Attended a webinar	8
Downloaded a guide	5
Clicked a link in an email	3

Negative points

Signal	Points
Competitor or job applicant	Remove from scoring
Personal email address (B2B only)	-15
Outside the regions you serve	-20
No activity in 90 days	-10

Score bands

Fit	Intent	Band	Action
30+	25+	A: hot	Call now, with a named owner and an alert
30+	Under 25	B: worth working	Add to an outbound sequence
Under 30	25+	C: curious	Qualify quickly by email or a short call
Under 30	Under 25	D: not now	Marketing nurture only, no rep time

A lead scoring example

Say you sell software to accounting firms with 10 to 50 staff, and three leads arrive on Monday morning.

Lead	Fit	Intent	Band	What happens
Managing partner, 25-person firm, requested a demo	45	30	A	Called within minutes
Operations manager, 15-person firm, visited pricing	38	10	B	Added to a sequence
Student on a Gmail address, downloaded two guides	-10	10	D	Nurture only

The point is the order. Your rep calls the partner first, even if the other two arrived earlier. That matters most for your A band, where speed to lead decides whether you reach the buyer before someone else does.

How to set this up in your CRM

Start in a spreadsheet if you have no CRM automation. Once the rules work, move them into your tools. HubSpot's scoring tool, for example, supports separate fit, engagement and combined scores, plus score decay, which lowers points as an event gets older. Check that your plan includes it.

Whichever tool you use, connect the score to something that happens automatically:

• Band A creates a task and alerts the named owner.
• Band B enrols the lead in a sequence, using a follow-up cadence that fits.
• Band C goes to a quick qualifying step.
• Band D stays in marketing, away from your reps.

A score that nobody acts on is just another CRM field.

Common lead scoring mistakes

• Mixing fit and intent into one number. Busy non-buyers end up at the top of the queue.
• Guessing the points in a meeting. Use deal data, even a small sample.
• Rewarding every click. Opens and page views inflate scores without showing intent.
• Never lowering scores. A guide downloaded eight months ago says little about today.
• Building too many rules. If a rep cannot explain a score, they will ignore it.
• Skipping sales input. Reps know which leads waste their time, so ask them first.

Who should build and run your scoring model?

The founder or sales lead should decide what a good lead is, because that is a commercial call. Building it is a different job: cleaning fields, enriching data, setting up the routing and fixing it when something breaks. That work suits a GTM engineer, and it is hard to keep up as a side task.

Running lead scoring with a dedicated remote hire

A scoring model only pays off when someone works the queue it creates. Through Zenstack, a full-time remote GTM engineer can build and maintain the scoring, enrichment and routing, while a remote SDR or appointment setter works the top band every day, in your hours and your CRM.

Rates start at $12 an hour, about $1,920 a month for a full-time hire, with recruiting, contracts and payroll handled by us. See our pricing, or tell us about your lead flow and we will work out which role to start with.